We compared the prophylactic efficacy of lithium, carbamazepine, and the combination in a double-blind, randomized, crossover design in 52 outpatients with bipolar disorder. Non-responders have a fourth year on lithium plus valproate and a fifth year on triple mood-stabilizer therapy. We have demonstrated the validity and reliability of the prospective life chart methodology in 27 patients and this is now being widely used as a major longitudinal assessment tool in the field. For evaluable patients: 13/42 (31%) failed to complete a full year on lithium due to lack of efficacy, and two dropped out due to side effects; 13/35 (37.1%) failed carbamazepine and ten dropped out due to side effects (9 had a rash); and 7/27 (25.9%) failed the combination. The percentage of the evaluable patients who had marked or moderate improvement on the Clinical Global Impressions (CGI) scale was 33.3% on lithium, 31.5% on carbamazepine, and 51.8% on the combination. The treatment response varied across different drugs for individual patients. Lithium was more effective than carbamazepine in the prophylaxis of mania. Rapid cycling patients did poorly on monotherapy (responding on lithium - 28.0%, on carbamazepine - 19.0%) but much better on the combination (53.3%). Having a first treatment by age 20 or earlier, and having no more than one lifetime hospitalization for mania predicted a positive response to lithium. Six of 18 evaluable patients were rated as having a good treatment response. Of 13 evaluable patients who had not responded or had treatment-limiting side effects on lithium, carbamazepine, or the combination, four were rated as having a good treatment response on val- proate plus lithium. Valproate provided a significant reduction in the percentage of time manic and the average severity of mania compared to baseline measures derived from the retrospective life chart. Thirty- three percent were rated as having a definite or likely or antidepressant-induced switch to mania that was most likely attributed to the natural course of illness. In addition, 22% of the patients experienced cycle acceleration in response to treatment with antidepressants. All of the data indicate the need for the development of new drugs and treatment regimens for this potentially life threatening recurrent illness. A new randomized trial of adjunct lamotrigine vs. gabapentin is planned. We have started a randomized antidepressant comparison of sertraline, bupropion, and venlafaxine for bipolar depression at one of the five Stanley Foundation Bipolar Network sites.